The Commission believes these proposals would further the purpose of the Rating Agency Act to improve the quality of NRSRO credit ratings by fostering accountability, transparency, and competition in the credit rating industry.\16\
\16\ See Senate Report, p. 2.

C. The Role of Credit Ratings in the Credit Market Turmoil

The growth in the origination of subprime loans began in the early 2000s.\17\ For example, Moody's reports that subprime loans amounted to $421 billion of the $3.038 trillion in mortgages originated in 2002 (14%) and $640 billion of the $2.886 trillion in mortgages originated in 2006 (22%).\18\ This growth was facilitated by steadily rising home values and a low interest rate environment.\19\ In addition, increases in the breadth of the credit risk transfer markets as a result of new investors willing to purchase credit based structured finance products provided an opportunity for lenders to originate subprime loans and then move them off their balance sheets by packaging and selling them through the securitization process to investors as subprime RMBS and CDOs.\20\ The investors in subprime RMBS and CDOs included domestic and foreign mutual funds, pension funds, hedge funds, banks, insurance companies, special investment vehicles, and state government operated funds.
\17\ See e.g., Statement of Sheila C. Bair, Chairman, Federal Deposit Insurance Corporation, before U.S. Senate Committee on Banking, Housing, and Urban Affairs (January 31, 2008) (``Bair January 31, 2008 Senate Statement''), p. 4.
\18\ According to Moody's, subprime mortgage loans represented $421 billion of $3.038 trillion total mortgage origination in 2002 and $640 billion of $2.886 trillion total mortgage origination in 2006. See A Short Guide to Subprime, Moody's, March 25, 2008, p. 1. \19\ See e.g., Dugan March 4, 2008 Senate Testimony, pp. 811. \20\ Id.

This ``originate to distribute'' business model created demand for residential
[[Page 36214]]
mortgage loans, including subprime loans. For example, according to Moody's, of the approximately $2.5 trillion worth of mortgage loans originated in 2006, $1.9 trillion were securitized into RMBS and approximately 25%, or $520 billion worth, of these loans were categorized as subprime.\21\ The demands of the loan securitization markets encouraged lenders to lower underwriting standards to maintain a steady volume of loans and to use less traditional products such as adjustable rate, negative amortization, and closedend second lien mortgages.\22\
\21\ Subprime Residential Mortgage Securitizations: Frequently Asked Questions, Moody's, April 19, 2007, p. 1.
\22\ See e.g., Bernanke February 28, 2008 Senate Testimony, p. 1; Dugan March 4, 2008 Senate Testimony, pp. 810.

1. The Creation of Subprime RMBS and CDOs

The creation of an RMBS begins by packaging a pool of mortgage loans, usually numbering in the thousands, and transferring them to a bankruptcy remote trust. The trust purchases the loan pool and becomes entitled to the interest and principal payments made by the borrowers. The trust finances the purchase of the loan pool through the issuance of RMBS. The monthly interest and principal payments from the loan pool are used to make monthly interest and principal payments to the investors in the RMBS.

The trust typically issues different classes of RMBS (known as ``tranches'') offering a sliding scale of coupon rates based on the level of credit protection afforded to the security. Credit protection is designed to shield the tranche securities from loss of interest and principal arising from defaults of the loans backing the RMBS. The degree of credit protection afforded a tranche security is known as its ``credit enhancement'' and is provided through several means. The primary source of credit enhancement is subordination, which creates a hierarchy of loss absorption among the tranche securities. For example, if a trust issued securities in 10 different tranches of securities, the first (or senior) tranche would have nine subordinate tranches, the next highest tranche would have eight subordinate tranches and so on down the capital structure. Losses of interest and principal experienced by the trust from delinquencies and defaults among loans in the pool are allocated first to the lowest tranche until its principal amount is exhausted and then to the next lowest tranche and so on up the capital structure. Consequently, the senior tranche would not incur any loss until the principal amounts from all the lower tranches have been exhausted through the absorption of losses from the underlying loans.

A second form of credit enhancement is overcollateralization, which is the amount that the principal balance of the mortgage pool underlying the trust exceeds the principal balance of the tranche securities issued by the trust. This excess principal creates an additional ``equity'' tranche below the lowest tranche security to absorb losses. In the example above, the equity tranche would sit below the 10th tranche security and protect it from the first losses experienced as a result of defaulting loans.

A third form of credit enhancement is excess spread, which consists of the amount by which the interest derived from the underlying loans in the aggregate exceeds interest payments due to investors in the tranche securities in the aggregate plus the administrative expenses of the trust such as fees due the loan servicer as well as premiums due on derivatives contracts and bond insurance. In other words, the excess spread is the amount that the monthly interest income from the pool of loans exceeds the weighted average interest due to the RMBS bondholders. This excess spread can be used to build up loss reserves or pay off delinquent interest payments due to a tranche security.

A fourth form of credit enhancement sometimes employed is bond insurance. When used, bond insurance is typically purchased only for the senior RMBS tranche.

The creation of a typical CDO is similar to that of an RMBS. A bankruptcy remote trust is created to hold the CDO's assets and issue its securities. The underlying assets, however, are generally debt securities rather than mortgage loans. The CDO trust uses the interest and principal payments from the approximately 200 underlying debt securities to make interest and principal payments to investors in the securities issued by the trust. The trust is structured to provide differing levels of credit enhancement to the securities it issues. Similar to RMBS, credit enhancement is provided through subordination, overcollateralization, excess spread, and bond insurance. In addition to the underlying assets, one significant difference between a CDO and an RMBS is that the CDO may be actively managed such that its underlying assets change over time, whereas the mortgage loan pool underlying an RMBS remains static for the most part.

In recent years, CDOs have been some of the largest purchasers of subprime RMBS and the drivers of demand for those securities. For example, according to Fitch, the average percentage of subprime RMBS in the collateral pools of CDOs it rated grew from 43.3% in 2003 to 71.3% in 2006.\23\ Generally, the CDOs holding subprime RMBS issued fell into one of two categories: High grade and mezzanine. High grade CDOs are generally defined as those that hold RMBS tranches with AAA, AA, or A credit ratings, whereas mezzanine CDOs are those that hold RMBS tranches rated predominantly BBB. Securities issued by mezzanine CDOs pay higher yields than those issued by high grade CDOs since the BBB rated RMBS underlying the mezzanine CDOs pay higher yields than the AAA to A rated RMBS underlying high grade CDOs. In addition to CDOs holding subprime RMBS, a market for CDOs holding other CDOs that held subprime RMBS developed in recent years. These debt instruments are known as ``CDOssquared.''
\23\ Rating Stability of FitchRated Global Cash Mezzanine Structured Finance CDOs with Exposure to U.S. Subprime RMBS, Fitch, April 2, 2007, p. 1.

As the market for mortgage related CDOs grew, CDO issuers began to use credit default swaps to replicate the performance of subprime RMBS and CDOs. In this case, rather than purchasing subprime RMBS or CDOs, the CDO entered into credit default swaps referencing subprime RMBS or CDOs, or indexes on RMBS. These CDOs, in some cases, are composed entirely of credit default swaps (``synthetic CDOs'') or a combination of credit default swaps and cash RMBS (``hybrid CDOs''). The use of credit default swaps allowed the CDO securities to be issued more quickly, since the issuer did not have to wait to accumulate actual RMBS for the underlying collateral pool.

2. Determining Credit Ratings for Subprime RMBS and CDOs

A key step in the process of creating and ultimately selling a subprime RMBS and CDO is the issuance of a credit rating for each of the tranches issued by the trust (with the exception of the most junior ``equity'' tranche). The credit rating for each rated tranche indicated the credit rating agency's view as to the creditworthiness of the debt instrument in terms of the likelihood that the issuer would default on its obligations to make interest and principal payments on the debt instrument.\24\ To varying degrees,
[[Page 36215]]
many investors rely on credit ratings in making the decision to purchase subprime RMBS or CDOs, particularly with respect to the senior AAA rated tranches. Some investors use the credit ratings to assess the risk of the debt instruments. In part, this may be due to the large number of debt instruments in the market and their complexity. Other investors use credit ratings to satisfy client investment mandates regarding the types of securities they can invest in or to satisfy regulatory requirements based on certain levels of credit ratings, or a combination of these conditions. Moreover, investors typically only have looked to ratings issued by Fitch, Moody's, and S&P, which causes the arrangers \25\ of the subprime RMBS and CDOs to use these three NRSROs to obtain credit ratings for the tranche securities they brought to market.
\24\ See, e.g., Inside the Ratings: What Credit Ratings Mean, Fitch, August 2007 (``Inside the Ratings''), p. 2; Testimony of Michael Kanef, Group Managing Director, Moody's Investors Service, Before the United States Senate Committee on Banking, Housing, and Urban Affairs (September 26, 2007) (``Kanef September 26, 2007 Senate Testimony''), p. 2; PrinciplesBased Rating Methodology For Global Structured Finance Securities, S&P, May 29, 2007, p. 3. Since credit ratings are issued for tranches of RMBS and CDOs
individually, rather than for the issuers of those tranches, the NRSRO credit ratings are estimates of the probability of default of each RMBS or CDO tranche as an independent instrument.
\25\ As bankruptcy remote standalone legal entities, RMBS and CDO trusts had no employees. Consequently, they relied on third parties to create and manage them. The term ``arranger'' is used herein to refer to the party that oversees the creation of the RMBS and CDO, which would include the process of obtaining credit ratings for the various tranches. Frequently, the arranger also served as the underwriter of the securities.

The procedures followed by these three NRSROs in developing ratings for subprime RMBS are generally similar. The arranger of the RMBS initiates the rating process by sending the credit rating agency a range of data on each of the subprime loans to be held by the trust (e.g., principal amount, geographic location of the property, credit history and FICO score of the borrower, ratio of the loan amount to the value of the property, and type of loan: First lien, second lien, primary residence, secondary residence), the proposed capital structure of the trust, and the proposed levels of credit enhancement to be provided to each RMBS tranche issued by the trust. Upon receipt of the information, the NRSRO assigns a lead analyst who is responsible for analyzing the loan pool, proposed capital structure, and proposed credit enhancement levels and, ultimately, for formulating a ratings recommendation for a rating committee composed of analysts and/or seniorlevel personnel not involved in the analytic process.

The next step in the ratings process is the development of predictions, based on a quantitative expected loss model and other qualitative factors, as to how many of the loans in the collateral pool would default under stresses of varying severity. This analysis also includes assumptions as to how much principal would be recovered after a defaulted loan is foreclosed. Each NRSRO generally uses between 40 and 60 specific credit characteristics to analyze each loan in the collateral pool of an RMBS in order to assess the potential future performance of the loan under various possible scenarios. These characteristics include the loan information described above as well as the amount of equity that the borrowers have in their homes, the amount of documentation provided by borrowers to verify their assets and/or income levels, and whether the borrowers intend to rent or occupy the homes.\26\
\26\ See, e.g., Kanef September 26, 2007, Senate Testimony, p. 7.

The purpose of this loss analysis is to determine how much credit enhancement a given tranche security would need for a particular category of credit rating. The severest stress test (i.e., the one that would result in the greatest number of defaults among the underlying loans) is run to determine the amount of credit enhancement required for an RMBS tranche issued by the trust to receive an AAA rating. For example, this test might result in an output that predicted that under the ``worst case'' scenario, 40 percent of the loans in the underlying pool would default and that after default the trust would recover only 50 percent of the principal amount of each loan in foreclosure. Consequently, to get an AAA rating, an RMBS tranche security issued by the trust would need credit enhancement sufficient to cover at least 20 percent of the principal amount of all the RMBS tranches issued by the trust. In other words, absent other forms of credit enhancement such as excess spread, at least 20 percent of the principal amount of the RMBS tranches issued by the trust, including the equity tranche, would have to be subordinate to the senior tranche and, therefore, obligated to absorb the losses resulting from 40% of the underlying loans defaulting.\27\ The next severest stress test is run to determine the amount of credit enhancement required of the AA tranche and so on down the capital structure. The lowest rated tranche (typically BB or B) is analyzed under a more benign market scenario. Consequently, its required level of credit enhancementtypically provided primarily or exclusively by a subordinate equity trancheis based on the number of loans expected to default in the normal course given the lowest possible level of macroeconomic stress.
\27\ To the extent that the RMBS included other forms of credit enhancement besides the subordination and overcollateralization provided in this example, e.g., excess spread, this 20 percent subordination figure would be reduced accordingly.

Following the determination of the level of credit enhancement required for each credit rating category, the next step in the ratings process is to check the proposed capital structure of the RMBS against these requirements. For example, if the proposed structure would create a senior RMBS tranche that had 18 percent of the capital structure subordinate to it (the other RMBS tranches, including, as applicable, an equity tranche), the analyst reviewing the transaction might conclude that based on the output of the loss model the senior tranche should be rated AA since it would need 20 percent subordination to receive an AAA credit rating. Additionally, the analyst could take other factors into consideration such as the quality of the loan servicer or the actual performance of similar pools of loans underlying other RMBS trusts to determine that in this case 18 percent subordination would be sufficient to support an AAA rating (to the extent these factors were not covered by the model).

Typically, if the analyst concludes that the capital structure of the RMBS did not support the desired ratingsin the example above, if it determined that 18 percent credit enhancement is insufficient for the desired AAA ratingthis preliminary conclusion would be conveyed to the arranger. The arranger could accept that determination and have the trust issue the securities with the proposed capital structure and the lower rating or adjust the structure to provide the requisite credit enhancement for the senior tranche to get the desired AAA rating (e.g., shift 2 percent of the principal amount of the senior tranche to a lower tranche or add or remove certain mortgages from the proposed asset pool). Generally, arrangers aim for the largest possible senior tranche, i.e., to provide the least amount of credit enhancement possible, since the senior trancheas the highest rated tranchepays the lowest coupon rate of the RMBS' tranches and, therefore, costs the arranger the least to fund.

The next step in the process is a cash flow analysis on the interest and principal expected to be received by the trust from the pool of subprime loans to determine whether it will be sufficient to pay the interest and principal due on each RMBS tranche issued by the trust. The NRSROs use quantitative cash flow
[[Page 36216]]
models that analyze the amount of principal and interest payments expected to be generated from the loan pool each month over the terms of the RMBS tranche securities under various stress scenarios. The outputs of this model are compared against the priority of payments (the ``waterfall'') to the RMBS tranches specified in the trust legal documents. The waterfall documentation could specify over
collateralization and excess spread triggers that, if breached, would reallocate principal and interest payments from lower tranches to higher tranches until the minimum levels of overcollateralization and excess spread were reestablished. Ultimately, the monthly principal and interest payments derived from the loan pool need to be enough to satisfy the monthly payments of principal and interest due by the trust to the investors in the RMBS tranches as well as to cover the administrative expenses of the trust.

In addition to expected loss and cash flow analysis, the analysts review the legal documentation of the trust to evaluate whether it is bankruptcy remote, i.e., isolated from the effects of any potential bankruptcy or insolvency of the arranger. They also review operational and administrative risk associated with the trust, using the results of periodic examinations of the principal parties involved in the issuance of the security, including the mortgage originators, the issuer of the security, the servicer of the mortgages in the loan pool, and the trustee.\28\ In assessing the servicer, for example, an NRSRO might review its past performance with respect to loan collection, billing, recordkeeping, and the treatment of delinquent loans.
\28\ Principal parties are not rated de novo in each RMBS transaction; rather, each NRSRO has its own procedures and schedules for reviewing those parties on a periodic basis in order to incorporate its assessment of those entities into the rating process.

Following these steps, the analyst develops a rating recommendation for each RMBS tranche, which then is presented to a rating committee composed of analysts and/or seniorlevel personnel not involved in the analytic process. The rating committee votes on the ratings for each tranche and usually approaches the arranger privately to notify it of the ratings decisions. In most cases, an arranger can appeal a rating decision, although the appeal is not always granted (and, if granted, may not necessarily result in any change in the rating decision). Final ratings decisions are published and subsequently monitored through surveillance processes. The NRSRO typically is paid only if the credit rating is issued, though sometimes it receives a breakup fee for the analytic work undertaken even if the credit rating is not issued.

The process for assigning ratings to subprime CDOs also involves a review of the creditworthiness of each tranche of the CDO. As with RMBS, the process centers on an examination of the pool of assets held by the trust and analysis of how they would perform individually and in correlation during various stress scenarios. However, this analysis is based primarily on the credit rating of each RMBS or CDO in the underlying pool or referenced through a credit default swap entered into by the CDO. In other words, the credit rating is the primary characteristic of the underlying debt instruments that the NRSROs take into consideration when performing their loss analysis. Hence, this review of the debt instruments in the collateral pool and the potential correlations among those securities does not ``look through'' those securities to their underlying asset pools. The analysis, consequently, generally only goes one level down to the credit ratings of the underlying instruments or reference securities.

CDOs collateralized by RMBS or by other CDOs often are actively managed. Consequently, there can be frequent changes to the composition of the cash assets (RMBS or CDOs), synthetic assets (credit default swaps), or combinations of cash and synthetic assets in the underlying pool. As a result, NRSRO ratings for managed CDOs are based not on the closing date composition of the pool but instead on covenanted limits for each potential type of asset that could be put in the pool. Typically, following a postclosing period in which no adjustments can be made to a CDO's collateral pool, the CDO's manager has a predetermined period of several years in which to adjust that asset pool through various sales and purchases pursuant to covenants set forth in the CDO's indenture. These covenants set limitations and requirements for the collateral pools of CDOs, often by establishing minimum and maximum concentrations for certain types of securities or certain ratings.

NRSROs use a CDO's indenture guidelines to run ``worstcase'' scenarios based on the various permutations of collateral permitted under the indenture. For example, an indenture might specify that a CDO's collateral pool must include between 10 and 20 percent AAArated subprime RMBS, with the remaining 80 to 90 percent composed of investmentgrade, but not AAA, subprime RMBS. In preparing a rating for that CDO, an NRSRO will run its models based on all possible collateral pools permissible under the indenture guidelines, placing the most weight on the results from the weakest potential pools (i.e., the minimum permissible amount, 10 percent, of AAArated securities and the lowestrated investment grade securities for the remaining 90 percent). As with RMBS ratings, the model results are then compared against the capital structure of the proposed CDO to confirm that the level of subordination, overcollateralization and excess spread available to each tranche provides the necessary amount of credit enhancement to sustain a particular rating.
3. The Downgrades in Credit Ratings of Subprime RMBS and CDOs

As noted above, the development of the credit risk transfer markets gave rise to an ``originate to distribute'' model whereby mortgage loans are originated with the intent to securitize them. Under this model, arrangers earn fees from originating, structuring, and underwriting RMBS and servicing the loans underlying the RMBS, as well as frequently a third set of fees from structuring, underwriting, and managing CDOs composed of RMBS. Moreover, the yields offered by subprime RMBS and CDO tranches (as compared to other types of similarly rated debt instruments) led to increased investor demand for these debt instruments. The originate to distribute model creates incentives for originating high volumes of mortgage loans while simultaneously reducing the incentives to maintain high underwriting standards for making such loans. The continued growth of the housing market through 2006, which led to increased competition among lenders, also contributed to looser subprime loan underwriting standards.\29\ \29\ See e.g., Dugan March 4, 2008 Senate Testimony, p. 10; Bernanke February 28, 2008 Senate Testimony, p. 1.

By mid2006, however, the steady rise in home prices that had fueled this growth in subprime lending came to an end as prices began to decline.\30\ Moreover, widespread areas of the country began to experience declines whereas, in the past, poor housing markets generally had been confined to distinct geographic areas.\31\ The downturn in the housing market has been accompanied by a marked increase
[[Page 36217]]
in delinquencies and defaults of subprime loans.\32\
\30\ See e.g., Id; Bair March 4, 2008 Senate Statement, pp. 58; Bair January 31, 2008 Senate Statement, p. 3.
\31\ See e.g., Bair January 31, 2008 Senate Statement, p. 3. \32\ Id.

The increases in delinquency and default rates have been concentrated in loans made in 2006 and 2007, which indicates that borrowers have been falling behind within months of the loans being made.\33\ For example, by the fourth quarter of 2006, the percentage of subprime loans underlying RMBS rated by Moody's that were in default within six months of the loans being made stood at 3.54 percent, nearly four times the average six month default rate of 0.90 percent between the first quarter of 2002 and the second quarter of 2005. Similarly, default rates for subprime loans within 12 months of the loans being made rose to 7.39 percent as compared to 2.00 percent for the period from the first quarter of 2002 through the second quarter of 2005.\34\ Figures released by S&P show similar deterioration in the performance of recent subprime loans.\35\ According to S&P, the serious delinquency rate \36\ for subprime loans underlying RMBS rated by S&P within twelve months of the initial rating was 4.97 percent of the current aggregate pool balance for subprime RMBS issued in 2005, 10.55 percent for subprime RMBS issued in 2006, and 15.19 percent for subprime RMBS issued in 2007.\37\
\33\ See e.g., Bair March 24, 2008 Senate Statement, p. 6 (``Serious delinquency rates on subprime mortgages securitized in 2006 are significantly higher than those for any of the previous three years.'').
\34\ Early Defaults Rise in Mortgage Securitizations: Updated Data Show Continued Deterioration, Moody's, September 19, 2007, pp. 34.
\35\ U.S. Subprime RMBS Performance Update: January 2008 Distribution Date, S&P, February 25, 2008, p. 1.
\36\ Defined as 90plus day delinquencies, foreclosures, and real estate owned. Id.

\37\ Id.

Along with the deterioration in the performance of subprime loans, there has been an increase in the losses incurred after the loans are foreclosed. According to S&P, the actual realized losses on loans underlying 2007 subprime RMBS after 12 months of seasoning were 65 percent higher than the losses recorded for RMBS issued in 2006 at the same level of seasoning.\38\

\38\ Id.

The rising delinquencies and defaults in subprime loans backing the RMBS rated by the NRSROs has exceeded the projections on which they based their initial ratings. Furthermore, the defaults and foreclosures on subprime loans have resulted in realizable losses to the lower RMBS tranches backed by the loans and, correspondingly, to the lower CDO tranches backed by those RMBS. As discussed above, the reduction in the amount of monthly principal and interest payments coming from the underlying pool of subprime loans or, in the case of a CDO, RMBS tranches or other CDO tranches is allocated to the tranches in ascending order. In addition to directly impairing the affected tranche, the lossesby reducing the principal amount of these tranchesdecreased the level of subordination protecting the more senior tranches. In other words, losses suffered by the junior tranches of an RMBS or CDO directly reduced the level of credit enhancementthe primary factor considered by NRSROs in rating tranched securities protecting the senior tranches of the instrument. These factors have caused the NRSROs to reevaluate, and in many cases downgrade, their ratings for these instruments.

The extensive use of subprime RMBS in the collateral pools of CDOs has led to similar levels of downgrade rates for those securities as well. Moreover, the use of subprime RMBS as reference securities for synthetic CDOs magnified the effect of RMBS downgrades on CDO ratings. Surveillance of CDO credit ratings has been complicated by the fact that the methodologies used by the NRSROs to rate them relied heavily on the credit rating of the underlying RMBS or CDOs. Consequently, to adjust the CDO rating, the NRSROs first have needed to complete their reviews of the ratings for the underlying RMBS or adjust their methodologies to sufficiently account for the anticipated poor performance of the RMBS.\43\ Ultimately, the NRSROs have downgraded a substantial number of CDO ratings.
\43\ For example, in November 2007, Fitch announced that in rating CDOs with asset pools which included subprime RMBS, it would adjust all subprime RMBS securities on Rating Watch Negative downwards by three categoriesor notches(six in the case of 2007 subprime RMBS rated BBB+ or lower) before factoring them into a re assessment of the CDO's rating. See Global Criteria for the Review of Structured Finance CDOs With Exposure to U.S. Subprime RMBS, Fitch, November 15, 2007, p. 4.

\47\ Id., p. 6.

The scope and magnitude of these downgrades has caused a loss of confidence among investors in the reliability of RMBS and CDO credit ratings issued by the NRSROs.\48\ This lack of confidence in the accuracy of NRSRO ratings has been a factor in the broader dislocation in the credit markets.\49\ For example, the complexity of assessing the risk of structured finance products and the lack of commonly accepted methods for measuring the risk has caused investors to leave the market, including the market for AAA instruments, particularly investors that had relied primarily on NRSRO credit ratings in assessing whether to purchase these instruments.\50\ This has had a significant impact on the liquidity of the market for these instruments.\51\
\48\ See, e.g., Dugan March 4, 2008 Senate Testimony, p. 13. \49\ Id., Bair March 4, 2008 Senate Statement, p. 7.
\50\ Id., Bernanke February 28, 2008 Senate Testimony, p. 3. \51\ See, e.g., Dugan March 4, 2008 Senate Testimony, p. 13; Bair January 31, 2008 Senate Testimony, pp. 34.

In the wake of these events, the NRSROs that rated subprime RMBS and CDOs have come under intense criticism and scrutiny. It has been suggested that changes may be needed to address the conflicts of interest inherent in the process of rating RMBS and CDOs.\52\ The NRSROs that have been the primary ratings providers for subprime RMBS and related CDOs each operate under an ``issuerpays'' model in which they are paid by the arranger to rate a proposed RMBS or CDO. The arranger has an economic interest in obtaining the highest credit rating possible for each security issued by the trust and the NRSRO has an economic interest in having the arranger select it to rate the next RMBS or CDO brought by the arranger to market. Observers have questioned whether, given the incentives created by this arrangement, the NRSROs are able to issue unbiased ratings, particularly as the volume of deals brought by certain arrangers increased in the mid 2000s.\53\ The above concerns are compounded by the arrangers' ability to ``ratings shop.'' Ratings shopping is the process by which an arranger will bring its proposed RMBS and CDO transaction to multiple NRSROs and choose, on a dealwide or tranchebytranche basis, which two (or in some cases one) to use based on the preliminary ratings of the NRSROs.
\52\ See, e.g., Opening Statement of Senator Richard C. Shelby for the Hearing of the U.S. Senate Committee on Banking, Housing, and Urban Affairs (September 26, 2007), pp. 12.
\53\ See, e.g., Testimony of Professor John C. Coffee, Jr., Adolf A. Berle Professor of Law, Columbia University Law School, before the U.S. Senate Committee on Banking, Housing, and Urban Affairs (September 26, 2007), pp. 45.

In addition, the interaction between the NRSRO and the arranger during the RMBS and CDO rating process has raised concerns that the NRSROs are rating products they designed (i.e., evaluating their own work).\54\ A corporate issuer is more constrained in how it can adjust in response to an NRSRO to improve its creditworthiness in order to obtain a higher rating. In the context of structured finance products, the arranger has much more flexibility to make adjustments to obtain a desired credit rating by, for example, changing the composition of the assets in the pool held by the trust or the subordination levels of the tranche securities issued by the trust. In fact, an arranger frequently will inform the NRSRO of the rating it wishes to obtain for each tranche and will choose an asset pool, trust structure, and credit enhancement levels based on its understanding of the NRSRO's quantitative and qualitative models. The credit analyst will use the expected loss and cash flow models to, in effect, check whether the proposed assets, trust structure and credit enhancement levels are sufficient to support the credit ratings desired by the arranger. \54\ See, e.g., Opening Statement of Senator Jack Reed for the Hearing of the U.S. Senate Committee on Banking, Housing, and Urban Affairs (September 26, 2007), pp. 12.

The NRSRO rules adopted by the Commission in June of 2007 preceded the full emergence of the credit market turmoil. The Commission, in light of its experience since the final rules became effective, is proposing amendments to those rules and a new rule with the goal of further enhancing the utility of NRSRO disclosure to investors, strengthening the integrity of the ratings process, and more effectively addressing the potential for conflicts of interest inherent in the ratings process for structured finance products.
II. Proposed Amendments

A. Amendments to Rule 17g5

The Commission adopted Rule 17g5, in part, pursuant to authority ``to prohibit, or require the management and disclosure of, any conflicts of interest relating to the issuance of credit ratings by an [NRSRO].'' \55\ The rule identifies a series of conflicts arising from the business of determining credit ratings. Under the rule, some of these conflicts must be disclosed and managed, while other specified conflicts are prohibited outright.
\55\ See Section 15E(h)(2) of the Exchange Act (15 U.S.C. 78o 7(h)(2)).

Paragraph (a) of Rule 17g5 prohibits an NRSRO from having a conflict identified in paragraph (b) of the rule unless the NRSRO discloses the type of conflict on Form NRSRO and establishes, maintains, and enforces procedures to manage it.\56\ Paragraph (b) identifies eight types of conflicts, which include being paid by issuers or underwriters to determine credit ratings with respect to securities or money market instruments they issue or underwrite \57\ or being paid by persons for subscriptions to receive or access credit ratings where such persons also may own investments or have entered into transactions that could be favorably or adversely impacted by a credit rating.\58\
\56\ 17 CFR 240.17g5(a).
\57\ 17 CFR 240.17g5(b)(1).

\58\ 17 CFR 240.17g5(b)(5).

Paragraph (c) of Rule 17g5 prohibits outright four types of conflicts of interest. Consequently, an NRSRO would violate the rule if it has the type of conflict described in paragraph (c) even if it disclosed the conflict and established procedures to manage it. In the Adopting Release, the Commission explained that these conflicts were prohibited because they would be difficult to manage given their potential to cause undue influence.\59\

\59\ Adopting Release, 72 FR at 33598.

The Commission is proposing to amend Rule 17g5 to require the disclosure and establishment of procedures to manage an additional conflict and to prohibit certain other conflicts outright, as described below.
[[Page 36219]]
1. Addressing the Particular Conflict Arising From Rating Structured Finance Products by Enhancing the Disclosure of Information Used in the Rating Process

a. The Proposed Amendment

The Commission is proposing to amend Rule 17g5 \60\ to add to the list of conflicts that must be disclosed and managed the additional conflict of repeatedly being paid by certain arrangers to rate structured finance products. This conflict is a subset of the broader conflict already identified in paragraph (b)(1) of Rule 17g5; namely, ``being paid by issuers and underwriters to determine credit ratings with respect to securities or money market instruments they issue or underwrite.'' \61\ In the case of structured finance products, the Commission preliminarily believes this ``issuer/underwriterpay'' conflict is particularly acute because certain arrangers of structured finance products repeatedly bring ratings business to the NRSROs.\62\ As sources of constant deal based revenue, some arrangers have the potential to exert greater undue influence on an NRSRO than, for example, a corporate issuer that may bring far less ratings business to the NRSRO.\63\ Consequently, the Commission is proposing amendments to Rule 17g5 that would require additional measures to address this particular type of ``issuer/underwriterpay'' conflict.
\60\ 17 CFR 240.17g5.
\61\ 17 CFR 240.17g5(b)(1). As the Commission noted when adopting Rule 17g5, the concern with conflict identified in paragraph (b)(1) ``is that an NRSRO may be influenced to issue a more favorable credit rating than warranted in order to obtain or retain the business of the issuer or underwriter.'' Adopting Release, 72 FR at 33595.
\62\ See e.g., Testimony of Professor John C. Coffee, Jr., Adolf A. Berle Professor of Law, Columbia University Law School, before the U.S. Senate Committee on Banking, Housing, and Urban Affairs (April 22, 2008) (``Coffee April 22, 2008 Senate Testimony''), pp. 46.

\63\ Id.

Specifically, the proposed amendment would redesignate paragraph (b)(9) of Rule 17g5 as paragraph (b)(10) and in new paragraph (b)(9) identify the following conflict: issuing or maintaining a credit rating for a security or money market instrument issued by an asset pool or as part of any assetbacked or mortgagebacked securities transaction that was paid for by the issuer, sponsor, or underwriter of the security or money market instrument. To address this conflict, proposed new paragraph (a)(3) would require that as a condition to the NRSRO rating a structured finance product the information provided to the NRSRO and used by the NRSRO in determining the credit rating would need to be disclosed through a means designed to provide reasonably broad dissemination of the information.\64\ The intent behind this disclosure is to create the opportunity for other NRSROs to use the information to rate the instrument as well. Any resulting ``unsolicited ratings'' could be used by market participants to evaluate the ratings issued by the NRSRO hired to rate the product and, in turn, potentially expose an NRSRO whose ratings were influenced by the desire to gain favor with the arranger in order to obtain more business.\65\
\64\ This proposed requirement would be in addition to the current requirements of paragraph (a) that an NRSRO disclose the type of conflict of interest in Exhibit 6 to Form NRSRO; and establish, maintain and enforce written policies and procedures to address and manage the conflict of interest. 17 CFR 240 17g5(a)(1) and (2).
\65\ As used herein, an ``unsolicited rating'' is one that is determined without the consent and/or payment of the obligor being rated or issuer, underwriter, or arranger of the securities being rated.

The proposed amendment would require the disclosure of information provided to an NRSRO by the ``issuer, underwriter, sponsor, depositor, or trustee.'' The Commission preliminarily believes that, taken together, these are the parties that provide all relevant information to the NRSRO to be used in the initial rating and rating monitoring processes. The Commission is not proposing to specify the partyNRSRO, arranger, issuer, depositor, or trusteethat would need to disclose the information. It may be that the issuer through the arranger and trustee would be in the best positions to disclose the information. In this case, in contracting with these parties to provide a rating for a structured finance product, the NRSRO could require a representation from them that the necessary information would be disclosed as required by the proposed rule. The Commission notes, however, that the proposed rule does not provide a safe harbor for an NRSRO arising from such a representation. Consequently, an NRSRO would violate the proposed rule if it issued a credit rating for a structured finance product where the information is not disclosed notwithstanding any representations from the arranger.

The goal of this proposed amendment is to promote the effective management of this conflict of interest, increase the transparency of the process for rating structured finance products, and foster competition by making it feasible for more market participants, in particular NRSROs that are not contracted by the arranger to issue a rating but still wish to do so, to perform credit analysis on the instrument and to monitor the instrument's creditworthiness. As noted above, by providing the opportunity for more NRSROs to determine credit ratings for structured finance products, this proposal is designed to increase the number of ratings extant for a given instrument and, in particular, promote the issuance of ratings by NRSROs that are not hired by the arranger. The goal would be to expose an NRSRO that was unduly influenced by the ``arrangerpay'' conflict into issuing higher than warranted ratings.\66\ An ancillary benefit would be that the proposal could make it easier for users of credit ratings to identify potentially inaccurate credit ratings and incompetent NRSROs. The proposal also is designed to make it more difficult for arrangers to exert influence on the NRSROs that they hire to determine ratings for structured finance products. Specifically, by opening up the rating process to greater scrutiny, the proposal is designed to make it easier for the hired NRSRO to resist pressure from the arranger by increasing the likelihood that any steps taken to inappropriately favor the arranger could be exposed to the market. Further, as noted above, an ancillary benefit of the proposal is that it could operate as a check on inaccuracy and incompetence.
\66\ The Commission notes that ``unsolicited'' ratings could be used to obtain business with arrangers by creating a track record of favorable ratings. The Commission believes the potential to expose such conduct would be equal to that of exposing an NRSRO influenced by the ``arrangerpay'' conflict insomuch as the paid for ratings (usually at least two) would be consistently lower than the ``unsolicited'' ratings.

To further these goals, the proposal would require the disclosure of the following information:

\68\ See proposed paragraph (a)(3)(ii) of Rule 17g5.

For the purposes of the proposed amendment, the Commission would consider only information that is taken into account in generating the credit rating or in performing surveillance to be ``used'' by the NRSRO in those contexts. This would exclude information about collateral pools (i.e., ``loan tapes'') provided by the arranger containing a mix of assets that is different than the composition of the final collateral pool upon which the credit rating is based. The proposed rule also would exclude from disclosure most, if not all,
communications between the NRSRO and the issuer, underwriter, sponsor, depositor, or trustee to the extent the communications do not contain information necessary for the NRSRO to determine an initial credit rating or perform surveillance on an existing credit rating.

The Commission recognizes that the NRSRO would define the information that it uses for purposes of generating credit ratings and, likely, would obtain representations from the arranger that the information is being disclosed as required under the rule. There is a potential that an NRSRO that uses relatively little information to generate credit ratings would be favored by arrangers to minimize the amount of information subject to the disclosure requirement. The Commission preliminarily believes that there is some degree of standardization as to the information used by NRSROs to rate structured finance products (e.g., loan level information, payment priorities among the issued tranched securities, and legal structure of the issuer). An NRSRO that requires less than the standard level of information would need to convince users of credit ratings, most notably investors, that its ratings process was credible. Otherwise, arrangers ultimately would not use the NRSRO since it would be more difficult to sell the structured finance products if they carried ratings that were not accepted by the marketplace. Nonetheless, the Commission, if this proposal is adopted, intends to monitor whether it results in a significant reduction in the information provided to NRSROs.

The timing and scope of the disclosures of the first set information described aboveinformation used in determining the initial credit ratingwould depend on the nature of the offering: public, private, or offshore. \69\ In an offering registered under the Securities Act of 1933 (15 U.S.C. 77a et seq.), the information would need to be disclosed on the date the underwriter and the issuer or depositor set the offering price of the securities being rated (the ``pricing date''). \70\ In offerings that are not registered under the Securities Act of 1933 (15 U.S.C. 77a et seq.), the information would need to be disclosed to investors in the offering and entities meeting the definition of ``credit rating agency'' in Section 3(a)(61) of the Exchange Act (which would include credit rating agencies registered, and not registered, as NRSROs) \71\ and on the pricing date and disclosed publicly on the first business day after the transaction closes.
\69\ See Sections II.A.1.b.iiii below for a broader discussion of the scope of the disclosures that would be required under the proposed amendments.
\70\ See proposed paragraph (a)(3)(i)(A) of Rule 17g5.

\71\ 15 U.S.C. 78c(a)(61).

The Commission is proposing the pricing date as the time of the first disclosures because it preliminarily believes that this is the earliest date upon which the asset pool and legal structure of the trust are settled on. Thus, the information that would be disclosed would reflect the actual characteristics of the securities to be issued and not, for example, preliminary assets pools with different compositions of loans. At the same time, the disclosure of the information before the securities are sold is designed to provide the opportunity for other credit rating agencies to use the information to develop ``unsolicited ratings'' for the tranche securities before they are purchased by investors. To the extent unsolicited ratings are issued, they would provide investors with a greater range of credit assessments and, in particular, assessments from credit rating agencies that are not subject to the ``arrangerpay'' conflict.

The Commission anticipates that the information that would need to be disclosed (i.e., the information used by the hired NRSRO to determine the initial rating) generally would include the
characteristics of the assets in the pool underlying the structured finance product and the legal documentation setting forth the capital structure of the trust, payment priorities with respect to the tranche securities issued by the trust (the waterfall), and all applicable covenants regarding the activities of the trust. For example, for an initial rating for an RMBS, this information generally w

FOR FURTHER INFORMATION CONTACT Michael A. Macchiaroli, Associate Director, at (202) 5515525; Thomas K. McGowan, Assistant Director, at (202) 5515521; Randall W. Roy, Branch Chief, at (202) 5515522; Joseph I. Levinson, Attorney, at (202) 5515598; Carrie A. O'Brien, Attorney, at (202) 5515640; Sheila D. Swartz, Special Counsel, at (202) 551 5545; Rose Russo Wells, Special Counsel, at (202) 5515527; Division of Trading and Markets, Securities and Exchange Commission, 100 F Street, NE., Washington, DC 205496628 or, with respect to questions involving the proposed amendments as they implicate the Securities Act of 1933, Kathy Hsu, Special Counsel, at (202) 5513306 or Eduardo Aleman, Special Counsel, at (202) 5513646; Division of Corporation Finance, Securities and Exchange Commission, 100 F Street, NE, Washington, DC 205493628.